What should an AI visibility measurement show?
A useful measurement records the buyer questions, the AI systems and settings used, the time of observation and the complete answers. It distinguishes brand mentions, citations and recommendations while showing how many planned answers completed. Keep crawler visits and customer referrals as separate evidence. A failed answer is missing information; it cannot tell you that the brand was absent.
Six signals, six different questions
Choose the signal that answers your question. A total can conceal the difference between a system accessing a page and a buyer being given a reason to choose a business.
What each observation can tell you
Scroll to read all columns.
| Signal | Observation | What to inspect |
|---|---|---|
| Crawl | A client requested a public page. | URL, response status, time and traffic classification. |
| Retrieval | A source was fetched or returned for a query. | Retained retrieval or grounding evidence, when available. |
| Mention | The answer names the brand or product. | Surrounding wording and whether the offering is described correctly. |
| Citation | The answer includes a source reference. | The cited page and the claim it supports. |
| Recommendation | The answer suggests the product for a need. | Strength, conditions and other options in the answer. |
| Referral | A person arrives from an attributable source. | The visit and the action the person takes. |
Sources: How Ready Scan builds an AI visibility scorecard; The Ready Scan report guide; AI Performance in Bing Webmaster Tools
Use questions that represent a buying need
A question naming your business mainly tests how a known subject is described. A question about a problem or selection need can show which options the answer introduces. Keep those purposes distinct. Use a small set of different needs rather than many rewrites of one question, and retain wording, language, relevant region, observation date and measured surface.
Make the need meaningful
“How can a small team turn a weekly meeting into assigned follow-up tasks?” gives a concrete job to evaluate. Add a location or existing tool only when it changes the decision.
Save the wording
An edited question is a different observation even when its topic sounds similar.
Keep the context
Different AI products and search configurations can produce different answers. A familiar logo alone does not define the measurement.
Sources: Buyer questions for AI visibility
Read the wording before assigning a label
A name in a list is a mention. A source link is a citation. A recommendation requires wording suggesting the product for the need. Retain enough surrounding text to review that interpretation.
Illustrative wording only; these are not measured provider answers.
Scroll to read all columns.
| Hypothetical wording | Interpretation |
|---|---|
| “ExampleBrand makes task-management software.” | A mention; the sentence does not suggest using it. |
| “See ExampleBrand’s documentation for the export format.” | A cited source; check which claim it supports. |
| “ExampleBrand is an option if you need shared task assignments.” | A qualified recommendation tied to a condition. |
| “I could not retrieve an answer.” | A failed observation, not brand absence. |
Sources: How Ready Scan builds an AI visibility scorecard; The Ready Scan report guide
Show coverage beside every rate
This arithmetic is illustrative, not a Ready Scan scan or market benchmark. Suppose eight answers were planned, six completed, three of those six mentioned the brand and two recommended it. Mention rate among completed answers is 3 ÷ 6 = 50%; recommendation rate is 2 ÷ 6, about 33%. Coverage is 6 ÷ 8 = 75%. Using 3 ÷ 8 for mention rate would quietly treat two unknown outcomes as absence. Show both the numerator and denominator.
Sources: How Ready Scan builds an AI visibility scorecard; The Ready Scan report guide
Follow each citation to its claim
A citation may point to your website, a review, a directory or another source. Record which page appeared and check whether it supports the nearby statement. A homepage link does not explain every assertion about a product. Useful evidence identifies the destination and the part of the answer it supports.
Sources: Sources behind AI answers; The Ready Scan report guide; AI visibility insights: intent, topics and citation share
Make a useful change from the evidence
Clarify a confused product description, add a missing supported capability, repair an inaccessible page or explain when the product fits. Preserve useful existing content. Missing evidence calls for recovering the observation or inspecting the source before prescribing a confident change.
Sources: How Ready Scan builds an AI visibility scorecard; Sources behind AI answers
How Ready Scan supports this workflow
Ready Scan lets an authorised user review website context and select one to four buyer questions before answer acquisition. The report keeps completed answers and sources, shows coverage and separates observed outcomes. Read the methodology and report guide for the current process. The score opens the conversation; the retained evidence helps you decide what the result means.
Sources: How Ready Scan builds an AI visibility scorecard; The Ready Scan report guide
Sources
- How Ready Scan builds an AI visibility scorecardReady Scan · Accessed Explore publications using this source
- The Ready Scan report guideReady Scan · Accessed Explore publications using this source
- Buyer questions for AI visibilityReady Scan · Accessed Explore publications using this source
- Sources behind AI answersReady Scan · Accessed Explore publications using this source
- AI Performance in Bing Webmaster ToolsMicrosoft Bing · Accessed Explore publications using this source
- AI visibility insights: intent, topics and citation shareMicrosoft Bing · Accessed Explore publications using this source
Read next
Check whether AI includes your brand.
Ask the buyer questions that matter and preserve the exact answers before making a plan.
Check my brand